A Compressive Sensing Model for Speeding Up Text Classification

Table 3

Accuracies of different classifiers driven by BOW and CS features on binary and multiclass classification datasets when SRM is Block DCT.

Classifier	BOW feature	Subrate R for CS feature
		0.1	0.2	0.3	0.4	0.5	0.6
Binary classification
SVM	0.7220	0.6975	0.7135	0.7285	0.7265	0.7290	0.7265
Decision tree	0.6235	0.6365	0.6395	0.6460	0.6355	0.6465	0.6485
AdaBoost	0.7060	0.7020	0.6975	0.7075	0.7035	0.7020	0.7110
KNN	0.6040	0.5955	0.6120	0.6200	0.6140	0.6145	0.6125
Naïve Bayes	0.7275	0.7035	0.7130	0.7125	0.7170	0.7200	0.7150
Avg.	0.6766	0.6670	0.6751	0.6829	0.6793	0.6824	0.6827
Multiclass classification
SVM	0.8732	0.8358	0.8651	0.8666	0.8712	0.8767	0.8803
Decision tree	0.8560	0.8454	0.8434	0.8510	0.8520	0.8525	0.8530
AdaBoost	0.7777	0.7535	0.7737	0.7732	0.7813	0.7808	0.7818
KNN	0.8252	0.8080	0.8146	0.8207	0.8242	0.8257	0.8252
Naïve Bayes	0.7737	0.7373	0.7404	0.7464	0.7429	0.7424	0.7454
Avg.	0.8212	0.7960	0.8074	0.8116	0.8143	0.8156	0.8171